This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Update Win32 from version 0.54 to 0.56
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
3f620621 21=for apidoc_section $string
78342678 22=for apidoc AmnU||Nullch
72d33970
FC
23Null character pointer. (No longer available when C<PERL_CORE> is
24defined.)
2307c6d0 25
3f620621 26=for apidoc_section $SV
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
51b56f5c
KW
31
32Below are signatures of functions from config.h which can't easily be gleaned
33from it, and are very unlikely to change
34
3f620621 35=for apidoc_section $signals
51b56f5c
KW
36=for apidoc Am|int|Sigsetjmp|jmp_buf env|int savesigs
37=for apidoc Am|void|Siglongjmp|jmp_buf env|int val
38
3f620621 39=for apidoc_section $filesystem
51b56f5c
KW
40=for apidoc Am|void *|FILE_ptr|FILE * f
41=for apidoc Am|Size_t|FILE_cnt|FILE * f
42=for apidoc Am|void *|FILE_base|FILE * f
b290de04 43=for apidoc Am|Size_t|FILE_bufsiz|FILE *f
51b56f5c 44
3f620621 45=for apidoc_section $string
b290de04
KW
46=for apidoc Amu|token|CAT2|token x|token y
47=for apidoc Amu|string|STRINGIFY|token x
51b56f5c 48
3f620621 49=for apidoc_section $numeric
b290de04 50=for apidoc Am|double|Drand01
51b56f5c
KW
51=for apidoc Am|void|seedDrand01|Rand_seed_t x
52=for apidoc Am|char *|Gconvert|double x|Size_t n|bool t|char * b
53
54=cut
954c1994
GS
55*/
56
24792b8d
NC
57# define Nullch Null(char*)
58# define Nullfp Null(PerlIO*)
59# define Nullsv Null(SV*)
60#endif
8d063cd8 61
641d3f0b
PP
62#ifdef TRUE
63#undef TRUE
64#endif
65#ifdef FALSE
66#undef FALSE
67#endif
68#define TRUE (1)
69#define FALSE (0)
70
df87895c 71/*
3f620621 72=for apidoc_section $SV
df87895c
KW
73=for apidoc Am|void *|MUTABLE_PTR|void * p
74=for apidoc_item |AV *|MUTABLE_AV|AV * p
75=for apidoc_item |CV *|MUTABLE_CV|CV * p
76=for apidoc_item |GV *|MUTABLE_GV|GV * p
77=for apidoc_item |HV *|MUTABLE_HV|HV * p
78=for apidoc_item |IO *|MUTABLE_IO|IO * p
79=for apidoc_item |SV *|MUTABLE_SV|SV * p
80
81The C<MUTABLE_I<*>>() macros cast pointers to the types shown, in such a way
82(compiler permitting) that casting away const-ness will give a warning;
83e.g.:
84
85 const SV *sv = ...;
86 AV *av1 = (AV*)sv; <== BAD: the const has been silently
87 cast away
88 AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
89
54ba56f3
KW
90C<MUTABLE_PTR> is the base macro used to derive new casts. The other
91already-built-in ones return pointers to what their names indicate.
df87895c
KW
92
93=cut
cf3f0ffb
DM
94 */
95
041c1a23 96#if defined(PERL_USE_GCC_BRACE_GROUPS)
6c2255e0 97# define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
b1bc3f34
NC
98#else
99# define MUTABLE_PTR(p) ((void *) (p))
100#endif
101
a062e10d 102#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 103#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 104#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 105#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 106#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 107#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 108
f789f6a4 109#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
110# include <stdbool.h>
111# ifndef HAS_BOOL
112# define HAS_BOOL 1
113# endif
114#endif
115
8e84507e 116/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
117 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
118 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 119 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
120 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
121 g++ can be identified by __GNUG__.
122 Andy Dougherty February 2000
5d94fbed 123*/
3609ea0d 124#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 125# ifndef PERL_BOOL_AS_CHAR
5d94fbed 126# ifndef HAS_BOOL
c1d22f6b 127# define HAS_BOOL 1
5d94fbed 128# endif
f789f6a4 129# endif
5d94fbed
AD
130#endif
131
132#ifndef HAS_BOOL
f789f6a4
FC
133# ifdef bool
134# undef bool
135# endif
70d5cb32 136# define bool char
c1d22f6b 137# define HAS_BOOL 1
a687059c 138#endif
0d3e774c 139
25ba28ce 140/*
3f620621 141=for apidoc_section $casting
25ba28ce
KW
142=for apidoc Am|bool|cBOOL|bool expr
143
144Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
145if C<bool> is defined as C<char>, for example, then the cast from C<int> is
146implementation-defined.
147
148C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
149
150=cut
151*/
18f5643b 152#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 153
46c6c7e2 154/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
155 * XXX Should really be a Configure probe, with HAS__FUNCTION__
156 * and FUNCTION__ as results.
157 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
158#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
159# define FUNCTION__ __func__
7adf2470 160#elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
07798b17 161# define FUNCTION__ ""
46c6c7e2 162#else
07798b17 163# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
164#endif
165
27d4fb96
PP
166/* XXX A note on the perl source internal type system. The
167 original intent was that I32 be *exactly* 32 bits.
168
169 Currently, we only guarantee that I32 is *at least* 32 bits.
170 Specifically, if int is 64 bits, then so is I32. (This is the case
171 for the Cray.) This has the advantage of meshing nicely with
172 standard library calls (where we pass an I32 and the library is
173 expecting an int), but the disadvantage that an I32 is not 32 bits.
174 Andy Dougherty August 1996
24fef2a7 175
dc45a647
MB
176 There is no guarantee that there is *any* integral type with
177 exactly 32 bits. It is perfectly legal for a system to have
178 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 179
dc45a647
MB
180 Similarly, there is no guarantee that I16 and U16 have exactly 16
181 bits.
693762b4 182
8e84507e
NIS
183 For dealing with issues that may arise from various 32/64-bit
184 systems, we will ask Configure to check out
8175356b 185
1604cfb0
MS
186 SHORTSIZE == sizeof(short)
187 INTSIZE == sizeof(int)
188 LONGSIZE == sizeof(long)
189 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
190 PTRSIZE == sizeof(void *)
191 DOUBLESIZE == sizeof(double)
192 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 193
27d4fb96
PP
194*/
195
69512466
JH
196#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
197# include <inttypes.h>
dd0eed91
JH
198# ifdef INT32_MIN_BROKEN
199# undef INT32_MIN
200# define INT32_MIN (-2147483647-1)
201# endif
202# ifdef INT64_MIN_BROKEN
203# undef INT64_MIN
204# define INT64_MIN (-9223372036854775807LL-1)
205# endif
69512466
JH
206#endif
207
8175356b
JH
208typedef I8TYPE I8;
209typedef U8TYPE U8;
210typedef I16TYPE I16;
211typedef U16TYPE U16;
212typedef I32TYPE I32;
213typedef U32TYPE U32;
16d89be8 214
74b807c7 215#ifdef QUADKIND
8175356b
JH
216typedef I64TYPE I64;
217typedef U64TYPE U64;
16d89be8 218#endif
8175356b 219
5ff3f7a4
GS
220/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
221 Please search CHAR_MAX in perl.h for further details. */
26e4a9a9
KW
222#ifdef UINT8_MAX
223# define U8_MAX UINT8_MAX
0e983133 224#else
26e4a9a9 225# define U8_MAX PERL_UCHAR_MAX
0e983133 226#endif
26e4a9a9
KW
227#ifdef UINT8_MIN
228# define U8_MIN UINT8_MIN
5ff3f7a4 229#else
26e4a9a9
KW
230# define U8_MIN PERL_UCHAR_MIN
231#endif
5ff3f7a4 232
26e4a9a9
KW
233#ifdef INT16_MAX
234# define I16_MAX INT16_MAX
79072805 235#else
26e4a9a9
KW
236# define I16_MAX PERL_SHORT_MAX
237#endif
238#ifdef INT16_MIN
239# define I16_MIN INT16_MIN
240#else
241# define I16_MIN PERL_SHORT_MIN
242#endif
243#ifdef UINT16_MAX
244# define U16_MAX UINT16_MAX
245#else
246# define U16_MAX PERL_USHORT_MAX
247#endif
248#ifdef UINT16_MIN
249# define U16_MIN UINT16_MIN
250#else
251# define U16_MIN PERL_USHORT_MIN
79072805
LW
252#endif
253
26e4a9a9
KW
254#ifdef INT32_MAX
255# define I32_MAX INT32_MAX
256#elif LONGSIZE > 4
257# define I32_MAX PERL_INT_MAX
258#else
259# define I32_MAX PERL_LONG_MAX
260#endif
261#ifdef INT32_MIN
262# define I32_MIN INT32_MIN
263#elif LONGSIZE > 4
264# define I32_MIN PERL_INT_MIN
265#else
266# define I32_MIN PERL_LONG_MIN
267#endif
268#ifdef UINT32_MAX
269# ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
270# define U32_MAX UINT_MAX
271# else
272# define U32_MAX 4294967295U
273# endif
274#elif LONGSIZE > 4
275# define U32_MAX PERL_UINT_MAX
276#else
277# define U32_MAX PERL_ULONG_MAX
278#endif
279#ifdef UINT32_MIN
280# define U32_MIN UINT32_MIN
281#elif LONGSIZE > 4
282# define U32_MIN PERL_UINT_MIN
283#else
284# define U32_MIN PERL_ULONG_MIN
5ff3f7a4
GS
285#endif
286
23332c7d
KW
287/*
288=for apidoc_section $integer
289=for apidoc Ay|| PERL_INT_FAST8_T
290=for apidoc_item PERL_INT_FAST16_T
291=for apidoc_item PERL_UINT_FAST8_T
292=for apidoc_item PERL_UINT_FAST16_T
293
294These are equivalent to the correspondingly-named C99 typedefs on platforms
295that have those; they evaluate to C<int> and C<unsigned int> on platforms that
296don't, so that you can portably take advantage of this C99 feature.
297
298=cut
299*/
247cee9f
KW
300# ifdef I_STDINT
301 typedef int_fast8_t PERL_INT_FAST8_T;
302 typedef uint_fast8_t PERL_UINT_FAST8_T;
303 typedef int_fast16_t PERL_INT_FAST16_T;
304 typedef uint_fast16_t PERL_UINT_FAST16_T;
305# else
306 typedef int PERL_INT_FAST8_T;
307 typedef unsigned int PERL_UINT_FAST8_T;
308 typedef int PERL_INT_FAST16_T;
309 typedef unsigned int PERL_UINT_FAST16_T;
310# endif
247cee9f 311
464decb6 312/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
ab350cbd
KW
313 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
314 * required to represent any possible unsigned number containing N bits.
315 * TYPE_DIGITS gives the number of decimal digits required to represent any
316 * possible unsigned number of type T. */
464decb6 317#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
318#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
319#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
320
88794300 321/* Unused by core; should be deprecated */
ff68c719 322#define Ctl(ch) ((ch) & 037)
8d063cd8 323
98fce2a4
KW
324#if defined(PERL_CORE) || defined(PERL_EXT)
325# ifndef MIN
326# define MIN(a,b) ((a) < (b) ? (a) : (b))
327# endif
328# ifndef MAX
329# define MAX(a,b) ((a) > (b) ? (a) : (b))
330# endif
331#endif
332
84ff4fa9
KW
333/* Returns a boolean as to whether the input unsigned number is a power of 2
334 * (2**0, 2**1, etc). In other words if it has just a single bit set.
335 * If not, subtracting 1 would leave the uppermost bit set, so the & would
336 * yield non-zero */
337#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 338# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
339#endif
340
d223e1ea 341/* Returns a mask with the lowest n bits set */
fae1e72b 342#define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
d223e1ea 343
1381ccb1
KW
344/* The largest unsigned number that will fit into n bits */
345#define nBIT_UMAX(n) nBIT_MASK(n)
346
8d9433eb 347/*
3f620621 348=for apidoc_section $directives
d23778e6 349=for apidoc Am||__ASSERT_|bool expr
8d9433eb
KW
350
351This is a helper macro to avoid preprocessor issues, replaced by nothing
352unless under DEBUGGING, where it expands to an assert of its argument,
353followed by a comma (hence the comma operator). If we just used a straight
354assert(), we would get a comma with nothing before it when not DEBUGGING.
355
356=cut
357
803e4935 358We also use empty definition under Coverity since the __ASSERT_
8d9433eb
KW
359checks often check for things that Really Cannot Happen, and Coverity
360detects that and gets all excited. */
3e94db23 361
e7ae132e
KW
362#if defined(DEBUGGING) && !defined(__COVERITY__) \
363 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
364# define __ASSERT_(statement) assert(statement),
365#else
366# define __ASSERT_(statement)
367#endif
368
3fe05580 369/*
3f620621 370=for apidoc_section $SV
3fe05580 371
3bb9fd01 372=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 373Like C<newSVpvn>, but takes a literal string instead of a
30a15352 374string/length pair.
3fe05580 375
3bb9fd01 376=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 377Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 378a string/length pair.
84bafc02 379
3bb9fd01 380=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 381Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 382a string/length pair and omits the hash parameter.
3fe05580 383
3bb9fd01 384=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 385Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 386of a string/length pair.
9dcc53ea 387
3bb9fd01 388=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 389Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 390a string/length pair.
9dcc53ea 391
3bb9fd01 392=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 393Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 394string/length pair.
3fe05580 395
3bb9fd01 396=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 397Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
398string/length pair.
399
3bb9fd01 400=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 401Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 402string/length pair.
3fe05580 403
3bb9fd01 404=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 405Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
406string/length pair.
407
3bb9fd01 408=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 409Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 410a string/length pair.
9dcc53ea 411
3f620621 412=for apidoc_section $string
3fe05580 413
3bb9fd01 414=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 415Like C<savepvn>, but takes a literal string instead of a
30a15352 416string/length pair.
3fe05580 417
3bb9fd01 418=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
419A version of C<savepvs()> which allocates the duplicate string in memory
420which is shared between threads.
421
3f620621 422=for apidoc_section $GV
3fe05580 423
3bb9fd01 424=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 425Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 426string/length pair.
3fe05580 427
3f620621 428=for apidoc_section $HV
3fe05580 429
3bb9fd01 430=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 431Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 432string/length pair.
3fe05580 433
3bb9fd01 434=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 435Like C<hv_store>, but takes a literal string instead of a
0c395ea5 436string/length pair
3fe05580
MHM
437and omits the hash parameter.
438
3f620621 439=for apidoc_section $lexer
510966aa 440
3bb9fd01 441=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 442
1568d13a 443Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 444a string/length pair.
510966aa 445
3fe05580
MHM
446=cut
447*/
448
a34e53fc 449/*
3f620621 450=for apidoc_section $string
2efa8cc7 451
a34e53fc
KW
452=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
453
454Returns two comma separated tokens of the input literal string, and its length.
455This is convenience macro which helps out in some API calls.
456Note that it can't be used as an argument to macros or functions that under
457some configurations might be macros, which means that it requires the full
458Perl_xxx(aTHX_ ...) form for any API calls where it's used.
459
460=cut
461*/
462
a34e53fc 463#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
464
465/* STR_WITH_LEN() shortcuts */
466#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
467#define newSVpvs_flags(str,flags) \
468 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 469#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
470#define sv_catpvs_flags(sv, str, flags) \
471 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
472#define sv_catpvs_nomg(sv, str) \
473 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
474#define sv_catpvs(sv, str) \
475 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
476#define sv_catpvs_mg(sv, str) \
477 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 478#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
479#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
480#define sv_setref_pvs(rv, classname, str) \
481 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 482#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
483#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
484#define gv_stashpvs(str, create) \
485 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
3752a9fe
KW
486
487#define gv_fetchpvs(namebeg, flags, sv_type) \
488 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type)
72659597 489#define gv_fetchpvn gv_fetchpvn_flags
9dcc53ea
Z
490#define sv_catxmlpvs(dsv, str, utf8) \
491 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 492
ba3a79e7 493
510966aa
Z
494#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
495
b96d8cd9 496#define get_cvs(str, flags) \
1604cfb0 497 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 498
9b6e9510 499/* internal helpers */
4a1bbd3d 500/* Transitional */
d67f622b
N
501#ifndef PERL_VERSION_MAJOR
502# define PERL_VERSION_MAJOR PERL_REVISION
4a1bbd3d
KW
503#else
504# undef PERL_REVISION /* We don't want code to be using these */
505#endif
d67f622b
N
506#ifndef PERL_VERSION_MINOR
507# define PERL_VERSION_MINOR PERL_VERSION
4a1bbd3d
KW
508#else
509# undef PERL_VERSION
510#endif
d67f622b
N
511#ifndef PERL_VERSION_PATCH
512# define PERL_VERSION_PATCH PERL_SUBVERSION
4a1bbd3d
KW
513#else
514# undef PERL_SUBVERSION
515#endif
516
517#define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
518 /* '10*' leaves room for things like alpha, beta, releases */ \
519 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
9b6e9510 520#define PERL_DECIMAL_VERSION_ \
d67f622b
N
521 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
522 PERL_VERSION_PATCH)
9b6e9510
KW
523
524/*
3f620621 525=for apidoc_section $versioning
4a1bbd3d 526=for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
9de44d19
KW
527=for apidoc_item PERL_VERSION_NE
528=for apidoc_item PERL_VERSION_LT
529=for apidoc_item PERL_VERSION_LE
530=for apidoc_item PERL_VERSION_GT
531=for apidoc_item PERL_VERSION_GE
9b6e9510 532
4a1bbd3d 533Returns whether or not the perl currently being compiled has the specified
9b6e9510
KW
534relationship to the perl given by the parameters. For example,
535
536 #if PERL_VERSION_GT(5,24,2)
537 code that will only be compiled on perls after v5.24.2
538 #else
539 fallback code
540 #endif
541
542Note that this is usable in making compile-time decisions
543
4a1bbd3d
KW
544You may use the special value '*' for the final number to mean ALL possible
545values for it. Thus,
546
547 #if PERL_VERSION_EQ(5,31,'*')
548
549means all perls in the 5.31 series. And
550
551 #if PERL_VERSION_NE(5,24,'*')
552
553means all perls EXCEPT 5.24 ones. And
554
555 #if PERL_VERSION_LE(5,9,'*')
556
557is effectively
558
559 #if PERL_VERSION_LT(5,10,0)
560
561This means you don't have to think so much when converting from the existing
562deprecated C<PERL_VERSION> to using this macro:
563
564 #if PERL_VERSION <= 9
565
566becomes
567
568 #if PERL_VERSION_LE(5,9,'*')
569
9b6e9510
KW
570=cut
571*/
572
4a1bbd3d
KW
573/* N.B. These don't work if the patch version is 42 or 92, as those are what
574 * '*' is in ASCII and EBCDIC respectively */
575# define PERL_VERSION_EQ(j,n,p) \
576 (((p) == '*') \
d67f622b
N
577 ? ( (j) == PERL_VERSION_MAJOR \
578 && (n) == PERL_VERSION_MINOR) \
4a1bbd3d
KW
579 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
580# define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
581
582# define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
583 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
584 (n), \
585 (((p) == '*') ? 0 : p)))
586# define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
587
588# define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
589 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
590 (((p) == '*') ? ((n)+1) : (n)), \
591 (((p) == '*') ? 0 : p)))
592# define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
9b6e9510 593
954c1994 594/*
3f620621 595=for apidoc_section $string
ccfc67b7 596
954c1994 597=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
598Test two C<NUL>-terminated strings to see if they are different. Returns true
599or false.
954c1994
GS
600
601=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
602Test two C<NUL>-terminated strings to see if they are equal. Returns true or
603false.
954c1994
GS
604
605=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
606Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
607second, C<s2>. Returns true or false.
954c1994
GS
608
609=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
610Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
611equal to the second, C<s2>. Returns true or false.
954c1994
GS
612
613=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
614Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
615the second, C<s2>. Returns true or false.
954c1994
GS
616
617=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
618Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
619or equal to the second, C<s2>. Returns true or false.
954c1994
GS
620
621=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
622Test two C<NUL>-terminated strings to see if they are different. The C<len>
623parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
624wrapper for C<strncmp>).
625
626=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
627Test two C<NUL>-terminated strings to see if they are equal. The C<len>
628parameter indicates the number of bytes to compare. Returns true or false. (A
629wrapper for C<strncmp>).
954c1994 630
bd18bd40
KW
631=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
632Test two buffers (which may contain embedded C<NUL> characters, to see if they
633are equal. The C<len> parameter indicates the number of bytes to compare.
b96bd7bf
KW
634Returns true or false. It is undefined behavior if either of the buffers
635doesn't contain at least C<len> bytes.
bd18bd40 636
3bb9fd01 637=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
638Like L</memEQ>, but the second string is a literal enclosed in double quotes,
639C<l1> gives the number of bytes in C<s1>.
b96bd7bf 640Returns true or false.
2d8eeddb 641
bd18bd40
KW
642=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
643Test two buffers (which may contain embedded C<NUL> characters, to see if they
644are not equal. The C<len> parameter indicates the number of bytes to compare.
b96bd7bf
KW
645Returns true or false. It is undefined behavior if either of the buffers
646doesn't contain at least C<len> bytes.
bd18bd40 647
3bb9fd01 648=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
649Like L</memNE>, but the second string is a literal enclosed in double quotes,
650C<l1> gives the number of bytes in C<s1>.
b96bd7bf 651Returns true or false.
2d8eeddb 652
4aada8b9
KW
653=for apidoc Am|bool|memCHRs|"list"|char c
654Returns the position of the first occurence of the byte C<c> in the literal
655string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
656treated as unsigned char. Thus this macro can be used to determine if C<c> is
657in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
658is C<NUL> (and the set doesn't include C<NUL>).
659
954c1994 660=cut
fc169e00
KW
661
662New macros should use the following conventions for their names (which are
663based on the underlying C library functions):
664
665 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
666
667 Each has two main parameters, string-like operands that are compared
668 against each other, as specified by the macro name. Some macros may
669 additionally have one or potentially even two length parameters. If a length
670 parameter applies to both string parameters, it will be positioned third;
671 otherwise any length parameter immediately follows the string parameter it
672 applies to.
673
674 If the prefix to the name is 'str', the string parameter is a pointer to a C
675 language string. Such a string does not contain embedded NUL bytes; its
676 length may be unknown, but can be calculated by C<strlen()>, since it is
677 terminated by a NUL, which isn't included in its length.
678
a3815e44 679 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
680 giving the maximum number of bytes to look at in each string. Even if both
681 strings are longer than the length parameter, those extra bytes will be
682 unexamined.
683
684 The 's' suffix means that the 2nd byte string parameter is a literal C
685 double-quoted string. Its length will automatically be calculated by the
686 macro, so no length parameter will ever be needed for it.
687
688 If the prefix is 'mem', the string parameters don't have to be C strings;
689 they may contain embedded NUL bytes, do not necessarily have a terminating
690 NUL, and their lengths can be known only through other means, which in
691 practice are additional parameter(s) passed to the function. All 'mem'
692 functions have at least one length parameter. Barring any 'l' or 's' suffix,
693 there is a single length parameter, in position 3, which applies to both
694 string parameters. The 's' suffix means, as described above, that the 2nd
695 string is a literal double-quoted C string (hence its length is calculated by
696 the macro, and the length parameter to the function applies just to the first
697 string parameter, and hence is positioned just after it). An 'l' suffix
698 means that the 2nd string parameter has its own length parameter, and the
699 signature will look like memFOOl(s1, l1, s2, l2).
700
701 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
702 substring of the 1st string. 'P' if present indicates that the substring
703 must be a "proper" one in tha mathematical sense that the first one must be
704 strictly larger than the 2nd.
705
954c1994
GS
706*/
707
62946e08 708
75400963
KW
709#define strNE(s1,s2) (strcmp(s1,s2) != 0)
710#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
711#define strLT(s1,s2) (strcmp(s1,s2) < 0)
712#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
713#define strGT(s1,s2) (strcmp(s1,s2) > 0)
714#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 715
75400963
KW
716#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
717#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 718
9d3980bc
KW
719#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
720#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 721
085b7534 722/* memEQ and memNE where second comparand is a string constant */
568a785a 723#define memEQs(s1, l, s2) \
777fa2cb 724 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 725#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 726
fdbb9a7c
KW
727/* Keep these private until we decide it was a good idea */
728#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
729
730#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
731
bdb7e3f0 732#define memBEGINs(s1, l, s2) \
30a6480c 733 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 734 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 735#define memBEGINPs(s1, l, s2) \
30a6480c 736 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 737 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 738#define memENDs(s1, l, s2) \
30a6480c 739 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 740 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 741#define memENDPs(s1, l, s2) \
30a6480c 742 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 743 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 744#endif /* End of making macros private */
bdb7e3f0 745
062b6850
KW
746#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
747#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
748#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
749#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
750
4aada8b9
KW
751#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
752
bbce6d69
PP
753/*
754 * Character classes.
755 *
756 * Unfortunately, the introduction of locales means that we
757 * can't trust isupper(), etc. to tell the truth. And when
758 * it comes to /\w+/ with tainting enabled, we *must* be able
759 * to trust our character classes.
760 *
761 * Therefore, the default tests in the text of Perl will be
762 * independent of locale. Any code that wants to depend on
763 * the current locale will use the tests that begin with "lc".
764 */
765
553cc101 766#ifdef USE_LOCALE
2304df62
AD
767# ifndef CTYPE256
768# define CTYPE256
769# endif
770#endif
771
954c1994 772/*
ccfc67b7 773
dcccc8ff 774=head1 Character classification
243effed
KW
775This section is about functions (really macros) that classify characters
776into types, such as punctuation versus alphabetic, etc. Most of these are
777analogous to regular expression character classes. (See
778L<perlrecharclass/POSIX Character Classes>.) There are several variants for
779each class. (Not all macros have all variants; each item below lists the
780ones valid for it.) None are affected by C<use bytes>, and only the ones
781with C<LC> in the name are affected by the current locale.
782
d713f9d9
KW
783The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
784treating it as a code point, and returns a boolean as to whether or not the
785character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
786ASCII character in the named class based on platform, Unicode, and Perl rules.
787If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 788
c98722a4 789Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
790with no suffix C<"_A">. This variant is used to emphasize by its name that
791only ASCII-range characters can return TRUE.
4b9734bf 792
d60679e1 793Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
794onto the platform. That is, the code points that are ASCII are unaffected,
795since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
796as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
797true when called with the code point 0xDF, which is a word character in both
4650c663 798ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
799If the input is a number that doesn't fit in an octet, FALSE is returned.
800(Perl's documentation uses a colloquial definition of Latin-1, to include all
801code points below 256.)
243effed 802
d713f9d9
KW
803Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
804inputs below 256, but if the code point is larger than 255, Unicode rules are
805used to determine if it is in the character class. For example,
d0da05db 806C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 807WITH MACRON in Unicode, and is a word character.
243effed 808
059703b0
KW
809Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
810but are used for UTF-8 encoded strings. The two forms are different names for
811the same thing. Each call to one of these classifies the first character of
812the string starting at C<p>. The second parameter, C<e>, points to anywhere in
813the string beyond the first character, up to one byte past the end of the
814entire string. Although both variants are identical, the suffix C<_safe> in
815one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
816provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
817C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
818some way, the program may croak, or the function may return FALSE, at the
819discretion of the implementation, and subject to change in future releases.
243effed 820
d713f9d9
KW
821Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
822but the result is based on the current locale, which is what C<LC> in the name
823stands for. If Perl can determine that the current locale is a UTF-8 locale,
824it uses the published Unicode rules; otherwise, it uses the C library function
825that gives the named classification. For example, C<isDIGIT_LC()> when not in
826a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
827returned if the input won't fit into an octet. On some platforms where the C
828library function is known to be defective, Perl changes its result to follow
829the POSIX standard's rules.
243effed 830
d713f9d9
KW
831Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
832than 256, but for larger ones it returns the Unicode classification of the code
833point.
243effed 834
059703b0
KW
835Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
836C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
837are different names for the same thing. Each call to one of these classifies
838the first character of the string starting at C<p>. The second parameter,
839C<e>, points to anywhere in the string beyond the first character, up to one
840byte past the end of the entire string. Although both variants are identical,
841the suffix C<_safe> in one name emphasizes that it will not attempt to read
842beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
843is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
844character is malformed in some way, the program may croak, or the function may
845return FALSE, at the discretion of the implementation, and subject to change in
846future releases.
ccfc67b7 847
6fdd32c3
KW
848=for apidoc Am|bool|isALPHA|UV ch
849=for apidoc_item ||isALPHA_A|UV ch
850=for apidoc_item ||isALPHA_L1|UV ch
851=for apidoc_item ||isALPHA_uvchr|UV ch
fa470d81
KW
852=for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
853=for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
6fdd32c3
KW
854=for apidoc_item ||isALPHA_LC|UV ch
855=for apidoc_item ||isALPHA_LC_uvchr|UV ch
fa470d81 856=for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
d713f9d9
KW
857Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
858analogous to C<m/[[:alpha:]]/>.
dcccc8ff 859See the L<top of this section|/Character classification> for an explanation of
fa470d81 860the variants.
8a58bdcf 861
f16858ed
KW
862=cut
863
f1460a66 864Here and below, we add the prototypes of these macros for downstream programs
f16858ed
KW
865that would be interested in them, such as Devel::PPPort
866
6fdd32c3
KW
867=for apidoc Am|bool|isALPHANUMERIC|UV ch
868=for apidoc_item ||isALPHANUMERIC_A|UV ch
869=for apidoc_item ||isALPHANUMERIC_L1|UV ch
870=for apidoc_item ||isALPHANUMERIC_uvchr|UV ch
fa470d81
KW
871=for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
872=for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
6fdd32c3
KW
873=for apidoc_item ||isALPHANUMERIC_LC|UV ch
874=for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch
fa470d81 875=for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
6fdd32c3
KW
876=for apidoc_item ||isALNUMC|UV ch
877=for apidoc_item ||isALNUMC_A|UV ch
878=for apidoc_item ||isALNUMC_L1|UV ch
879=for apidoc_item ||isALNUMC_LC|UV ch
880=for apidoc_item ||isALNUMC_LC_uvchr|UV ch
d713f9d9
KW
881Returns a boolean indicating whether the specified character is one of
882C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff 883See the L<top of this section|/Character classification> for an explanation of
fa470d81 884the variants.
15861f94 885
255b632a
KW
886A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
887this corresponds to the C language alphanumeric definition). Also
888there are the variants
889C<isALNUMC_A>, C<isALNUMC_L1>
890C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
891
6fdd32c3
KW
892=for apidoc Am|bool|isASCII|UV ch
893=for apidoc_item ||isASCII_A|UV ch
894=for apidoc_item ||isASCII_L1|UV ch
895=for apidoc_item ||isASCII_uvchr|UV ch
fa470d81
KW
896=for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
897=for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
6fdd32c3
KW
898=for apidoc_item ||isASCII_LC|UV ch
899=for apidoc_item ||isASCII_LC_uvchr|UV ch
fa470d81 900=for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
8a58bdcf 901Returns a boolean indicating whether the specified character is one of the 128
243effed 902characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 903On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
904character corresponds to an ASCII character. Variants C<isASCII_A()> and
905C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff 906See the L<top of this section|/Character classification> for an explanation of
fa470d81 907the variants.
059703b0
KW
908Note, however, that some platforms do not have the C library routine
909C<isascii()>. In these cases, the variants whose names contain C<LC> are the
910same as the corresponding ones without.
243effed 911
d98532ea
KW
912Also note, that because all ASCII characters are UTF-8 invariant (meaning they
913have the exact same representation (always a single byte) whether encoded in
914UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
915byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
916C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 917
6fdd32c3
KW
918=for apidoc Am|bool|isBLANK|UV ch
919=for apidoc_item ||isBLANK_A|UV ch
920=for apidoc_item ||isBLANK_L1|UV ch
921=for apidoc_item ||isBLANK_uvchr|UV ch
fa470d81
KW
922=for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
923=for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
6fdd32c3
KW
924=for apidoc_item ||isBLANK_LC|UV ch
925=for apidoc_item ||isBLANK_LC_uvchr|UV ch
fa470d81 926=for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
243effed 927Returns a boolean indicating whether the specified character is a
6aff1f14 928character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff 929See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
930the variants.
931Note,
da8c1a98
KW
932however, that some platforms do not have the C library routine
933C<isblank()>. In these cases, the variants whose names contain C<LC> are
934the same as the corresponding ones without.
243effed 935
6fdd32c3
KW
936=for apidoc Am|bool|isCNTRL|UV ch
937=for apidoc_item ||isCNTRL_A|UV ch
938=for apidoc_item ||isCNTRL_L1|UV ch
939=for apidoc_item ||isCNTRL_uvchr|UV ch
fa470d81
KW
940=for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
941=for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
6fdd32c3
KW
942=for apidoc_item ||isCNTRL_LC|UV ch
943=for apidoc_item ||isCNTRL_LC_uvchr|UV ch
fa470d81 944=for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
f16858ed 945
243effed 946Returns a boolean indicating whether the specified character is a
6aff1f14 947control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff 948See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
949the variants.
950On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
951
6fdd32c3
KW
952=for apidoc Am|bool|isDIGIT|UV ch
953=for apidoc_item ||isDIGIT_A|UV ch
954=for apidoc_item ||isDIGIT_L1|UV ch
955=for apidoc_item ||isDIGIT_uvchr|UV ch
fa470d81
KW
956=for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
957=for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
6fdd32c3
KW
958=for apidoc_item ||isDIGIT_LC|UV ch
959=for apidoc_item ||isDIGIT_LC_uvchr|UV ch
fa470d81
KW
960=for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
961
2787a470 962Returns a boolean indicating whether the specified character is a
6aff1f14 963digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 964Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff 965See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
966the variants.
967
6fdd32c3
KW
968=for apidoc Am|bool|isGRAPH|UV ch
969=for apidoc_item ||isGRAPH_A|UV ch
970=for apidoc_item ||isGRAPH_L1|UV ch
971=for apidoc_item ||isGRAPH_uvchr|UV ch
fa470d81
KW
972=for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
973=for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
6fdd32c3
KW
974=for apidoc_item ||isGRAPH_LC|UV ch
975=for apidoc_item ||isGRAPH_LC_uvchr|UV ch
fa470d81 976=for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
243effed 977Returns a boolean indicating whether the specified character is a
6aff1f14 978graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 979See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
980the variants.
981
6fdd32c3
KW
982=for apidoc Am|bool|isLOWER|UV ch
983=for apidoc_item ||isLOWER_A|UV ch
984=for apidoc_item ||isLOWER_L1|UV ch
985=for apidoc_item ||isLOWER_uvchr|UV ch
fa470d81
KW
986=for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
987=for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
6fdd32c3
KW
988=for apidoc_item ||isLOWER_LC|UV ch
989=for apidoc_item ||isLOWER_LC_uvchr|UV ch
fa470d81 990=for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
2787a470 991Returns a boolean indicating whether the specified character is a
6aff1f14 992lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff 993See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
994the variants
995
6fdd32c3
KW
996=for apidoc Am|bool|isOCTAL|UV ch
997=for apidoc_item ||isOCTAL_A|UV ch
998=for apidoc_item ||isOCTAL_L1|UV ch
2787a470 999Returns a boolean indicating whether the specified character is an
6aff1f14 1000octal digit, [0-7].
243effed
KW
1001The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
1002C<isOCTAL>.
1003
6fdd32c3
KW
1004=for apidoc Am|bool|isPUNCT|UV ch
1005=for apidoc_item ||isPUNCT_A|UV ch
1006=for apidoc_item ||isPUNCT_L1|UV ch
1007=for apidoc_item ||isPUNCT_uvchr|UV ch
fa470d81
KW
1008=for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
1009=for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1010=for apidoc_item ||isPUNCT_LC|UV ch
1011=for apidoc_item ||isPUNCT_LC_uvchr|UV ch
fa470d81 1012=for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
243effed 1013Returns a boolean indicating whether the specified character is a
6aff1f14
KW
1014punctuation character, analogous to C<m/[[:punct:]]/>.
1015Note that the definition of what is punctuation isn't as
243effed
KW
1016straightforward as one might desire. See L<perlrecharclass/POSIX Character
1017Classes> for details.
dcccc8ff 1018See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1019the variants.
1020
6fdd32c3
KW
1021=for apidoc Am|bool|isSPACE|UV ch
1022=for apidoc_item ||isSPACE_A|UV ch
1023=for apidoc_item ||isSPACE_L1|UV ch
1024=for apidoc_item ||isSPACE_uvchr|UV ch
fa470d81
KW
1025=for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
1026=for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
6fdd32c3
KW
1027=for apidoc_item ||isSPACE_LC|UV ch
1028=for apidoc_item ||isSPACE_LC_uvchr|UV ch
fa470d81 1029=for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
2787a470 1030Returns a boolean indicating whether the specified character is a
6aff1f14 1031whitespace character. This is analogous
398d098a 1032to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 1033this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
1034locale forms of this macro (the ones with C<LC> in their names) matched
1035precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1036in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1037(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff 1038See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1039the variants.
1040
6fdd32c3
KW
1041=for apidoc Am|bool|isPSXSPC|UV ch
1042=for apidoc_item ||isPSXSPC_A|UV ch
1043=for apidoc_item ||isPSXSPC_L1|UV ch
1044=for apidoc_item ||isPSXSPC_uvchr|UV ch
fa470d81
KW
1045=for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1046=for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
6fdd32c3
KW
1047=for apidoc_item ||isPSXSPC_LC|UV ch
1048=for apidoc_item ||isPSXSPC_LC_uvchr|UV ch
fa470d81 1049=for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
398d098a 1050(short for Posix Space)
779cf272
KW
1051Starting in 5.18, this is identical in all its forms to the
1052corresponding C<isSPACE()> macros.
398d098a
KW
1053The locale forms of this macro are identical to their corresponding
1054C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1055non-locale forms differ from their C<isSPACE()> forms only in that the
1056C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1057Otherwise they are identical. Thus this macro is analogous to what
1058C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 1059See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1060the variants.
1061
6fdd32c3
KW
1062=for apidoc Am|bool|isUPPER|UV ch
1063=for apidoc_item ||isUPPER_A|UV ch
1064=for apidoc_item ||isUPPER_L1|UV ch
1065=for apidoc_item ||isUPPER_uvchr|UV ch
fa470d81
KW
1066=for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1067=for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
6fdd32c3
KW
1068=for apidoc_item ||isUPPER_LC|UV ch
1069=for apidoc_item ||isUPPER_LC_uvchr|UV ch
fa470d81 1070=for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
2787a470 1071Returns a boolean indicating whether the specified character is an
6aff1f14 1072uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 1073See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1074the variants.
1075
6fdd32c3
KW
1076=for apidoc Am|bool|isPRINT|UV ch
1077=for apidoc_item ||isPRINT_A|UV ch
1078=for apidoc_item ||isPRINT_L1|UV ch
1079=for apidoc_item ||isPRINT_uvchr|UV ch
fa470d81
KW
1080=for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1081=for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1082=for apidoc_item ||isPRINT_LC|UV ch
1083=for apidoc_item ||isPRINT_LC_uvchr|UV ch
fa470d81 1084=for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
8eea39dd 1085Returns a boolean indicating whether the specified character is a
6aff1f14 1086printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff 1087See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1088the variants.
1089
6fdd32c3
KW
1090=for apidoc Am|bool|isWORDCHAR|UV ch
1091=for apidoc_item ||isWORDCHAR_A|UV ch
1092=for apidoc_item ||isWORDCHAR_L1|UV ch
1093=for apidoc_item ||isWORDCHAR_uvchr|UV ch
fa470d81
KW
1094=for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1095=for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
6fdd32c3
KW
1096=for apidoc_item ||isWORDCHAR_LC|UV ch
1097=for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch
fa470d81 1098=for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
6fdd32c3
KW
1099=for apidoc_item ||isALNUM|UV ch
1100=for apidoc_item ||isALNUM_A|UV ch
1101=for apidoc_item ||isALNUM_LC|UV ch
1102=for apidoc_item ||isALNUM_LC_uvchr|UV ch
243effed
KW
1103Returns a boolean indicating whether the specified character is a character
1104that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1105in a regular expression. A word character is an alphabetic character, a
1106decimal digit, a connecting punctuation character (such as an underscore), or
1107a "mark" character that attaches to one of those (like some sort of accent).
1108C<isALNUM()> is a synonym provided for backward compatibility, even though a
1109word character includes more than the standard C language meaning of
1110alphanumeric.
dcccc8ff 1111See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1112the variants.
1113C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1114C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1115C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1116include the platform's native underscore.
1117
6fdd32c3
KW
1118=for apidoc Am|bool|isXDIGIT|UV ch
1119=for apidoc_item ||isXDIGIT_A|UV ch
1120=for apidoc_item ||isXDIGIT_L1|UV ch
1121=for apidoc_item ||isXDIGIT_uvchr|UV ch
fa470d81
KW
1122=for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1123=for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1124=for apidoc_item ||isXDIGIT_LC|UV ch
1125=for apidoc_item ||isXDIGIT_LC_uvchr|UV ch
fa470d81 1126=for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
8a58bdcf 1127Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1128digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1129and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff 1130See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1131the variants.
1132
6fdd32c3
KW
1133=for apidoc Am|bool|isIDFIRST|UV ch
1134=for apidoc_item ||isIDFIRST_A|UV ch
1135=for apidoc_item ||isIDFIRST_L1|UV ch
1136=for apidoc_item ||isIDFIRST_uvchr|UV ch
fa470d81
KW
1137=for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1138=for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
6fdd32c3
KW
1139=for apidoc_item ||isIDFIRST_LC|UV ch
1140=for apidoc_item ||isIDFIRST_LC_uvchr|UV ch
fa470d81 1141=for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
3c3ecf18
KW
1142Returns a boolean indicating whether the specified character can be the first
1143character of an identifier. This is very close to, but not quite the same as
1144the official Unicode property C<XID_Start>. The difference is that this
1145returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff 1146See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1147the variants.
1148
6fdd32c3
KW
1149=for apidoc Am|bool|isIDCONT|UV ch
1150=for apidoc_item ||isIDCONT_A|UV ch
1151=for apidoc_item ||isIDCONT_L1|UV ch
1152=for apidoc_item ||isIDCONT_uvchr|UV ch
fa470d81
KW
1153=for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1154=for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1155=for apidoc_item ||isIDCONT_LC|UV ch
1156=for apidoc_item ||isIDCONT_LC_uvchr|UV ch
fa470d81 1157=for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
3c3ecf18
KW
1158Returns a boolean indicating whether the specified character can be the
1159second or succeeding character of an identifier. This is very close to, but
1160not quite the same as the official Unicode property C<XID_Continue>. The
1161difference is that this returns true only if the input character also matches
dcccc8ff 1162L</isWORDCHAR>. See the L<top of this section|/Character classification> for
fa470d81 1163an explanation of the variants.
f16858ed 1164
3f620621 1165=for apidoc_section $numeric
8eea39dd 1166
95a59cab 1167=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1168Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1169Behaviour is only well defined when isXDIGIT(*str) is true.
1170
e7c1e6c1 1171=head1 Character case changing
21da7284
KW
1172Perl uses "full" Unicode case mappings. This means that converting a single
1173character to another case may result in a sequence of more than one character.
1174For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1175character sequence C<SS>. This presents some complications The lowercase of
1176all characters in the range 0..255 is a single character, and thus
1177C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1178return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1179an API that does allow every possible legal result to be returned.) Likewise
1180no other function that is crippled by not being able to give the correct
1181results for the full range of possible inputs has been implemented here.
e7c1e6c1 1182
d713f9d9 1183=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1184Converts the specified character to uppercase. If the input is anything but an
1185ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1186C<toUPPER_A> is equivalent.
954c1994 1187
d0da05db
KW
1188=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1189Converts the code point C<cp> to its uppercase version, and
1190stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1191point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1192that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1193bytes since the uppercase version may be longer than the original character.
1194
1195The first code point of the uppercased version is returned
21da7284
KW
1196(but note, as explained at L<the top of this section|/Character case
1197changing>, that there may be more.)
1f607577 1198
059703b0 1199=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1200=for apidoc_item toUPPER_utf8_safe
a239b1e2
KW
1201Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1202extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1203stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1204that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1205bytes since the uppercase version may be longer than the original character.
1206
1207The first code point of the uppercased version is returned
21da7284
KW
1208(but note, as explained at L<the top of this section|/Character case
1209changing>, that there may be more).
1f607577 1210
059703b0
KW
1211It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1212S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1213the UTF-8 for the input character is malformed in some way, the program may
1214croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1215of the implementation, and subject to change in future releases.
a239b1e2 1216
912c8141 1217C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1f607577 1218
25200305
KW
1219=for apidoc Am|U8|toFOLD|U8 ch
1220Converts the specified character to foldcase. If the input is anything but an
1221ASCII uppercase character, that input character itself is returned. Variant
1222C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1223Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1224
d0da05db
KW
1225=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1226Converts the code point C<cp> to its foldcase version, and
1227stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1228point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1229that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1230bytes since the foldcase version may be longer than the original character.
1231
1232The first code point of the foldcased version is returned
21da7284
KW
1233(but note, as explained at L<the top of this section|/Character case
1234changing>, that there may be more).
1f607577 1235
059703b0 1236=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1237=for apidoc_item toFOLD_utf8_safe
a239b1e2
KW
1238Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1239extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1240stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1241that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1242bytes since the foldcase version may be longer than the original character.
1243
1244The first code point of the foldcased version is returned
21da7284
KW
1245(but note, as explained at L<the top of this section|/Character case
1246changing>, that there may be more).
1f607577 1247
059703b0 1248It will not attempt
a239b1e2
KW
1249to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1250true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1251input character is malformed in some way, the program may croak, or the
1252function may return the REPLACEMENT CHARACTER, at the discretion of the
1253implementation, and subject to change in future releases.
1254
912c8141 1255C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1f607577 1256
3cb048e5
KW
1257=for apidoc Am|UV|toLOWER|UV cp
1258=for apidoc_item |UV|toLOWER_A|UV cp
1259=for apidoc_item |UV|toLOWER_L1|UV cp
1260=for apidoc_item |UV|toLOWER_LATIN1|UV cp
1261=for apidoc_item |UV|toLOWER_LC|UV cp
1262=for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1263=for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1264=for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1265
1266These all return the lowercase of a character. The differences are what domain
1267they operate on, and whether the input is specified as a code point (those
1268forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1269case, the code point to use is the first one in the buffer of UTF-8 encoded
1270code points, delineated by the arguments S<C<p .. e - 1>>.
1271
1272C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the
1273lowercase of any uppercase ASCII-range code point. All other inputs are
1274returned unchanged. Since these are macros, the input type may be any integral
1275one, and the output will occupy the same number of bits as the input.
1276
1277C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave
1278identically as C<toLOWER> for ASCII-range input. But additionally will return
1279the lowercase of any uppercase code point in the entire 0..255 range, assuming
1280a Latin-1 encoding (or the EBCDIC equivalent on such platforms).
1281
1282C<toLOWER_LC> returns the lowercase of the input code point according to the
1283rules of the current POSIX locale. Input code points outside the range 0..255
1284are returned unchanged.
1285
1286C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return
1287value is identical to that of C<toLOWER_L1> for input code points in the 0..255
1288range. The lowercase of the vast majority of Unicode code points is the same
1289as the code point itself. For these, and for code points above the legal
1290Unicode maximum, this returns the input code point unchanged. It additionally
1291stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1292length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1293contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1294
1295NOTE: the lowercase of a code point may be more than one code point. The
1296return value of this function is only the first of these. The entire lowercase
1297is returned in C<s>. To determine if the result is more than a single code
1298point, you can do something like this:
1299
1300 uc = toLOWER_uvchr(cp, s, &len);
1301 if (len > UTF8SKIP(s)) { is multiple code points }
1302 else { is a single code point }
1303
1304C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only
1305difference between these and C<toLOWER_uvchr> is that the source for these is
1306encoded in UTF-8, instead of being a code point. It is passed as a buffer
1307starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1308buffer may certainly contain more than one code point; but only the first one
1309(up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1310malformed in some way, the program may croak, or the function may return the
1311REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1312change in future releases.
1f607577 1313
25200305
KW
1314=for apidoc Am|U8|toTITLE|U8 ch
1315Converts the specified character to titlecase. If the input is anything but an
1316ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1317C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1318range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1319not a concept used in locale handling, so there is no functionality for that.)
25200305 1320
d0da05db
KW
1321=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1322Converts the code point C<cp> to its titlecase version, and
1323stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1324point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1325that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1326bytes since the titlecase version may be longer than the original character.
1327
1328The first code point of the titlecased version is returned
21da7284
KW
1329(but note, as explained at L<the top of this section|/Character case
1330changing>, that there may be more).
1f607577 1331
059703b0 1332=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141
KW
1333=for apidoc_item toTITLE_utf8_safe
1334Convert the first UTF-8 encoded character in the sequence starting at C<p> and
a239b1e2 1335extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1336stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1337that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1338bytes since the titlecase version may be longer than the original character.
1339
1340The first code point of the titlecased version is returned
21da7284
KW
1341(but note, as explained at L<the top of this section|/Character case
1342changing>, that there may be more).
1f607577 1343
059703b0 1344It will not attempt
a239b1e2
KW
1345to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1346true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1347input character is malformed in some way, the program may croak, or the
1348function may return the REPLACEMENT CHARACTER, at the discretion of the
1349implementation, and subject to change in future releases.
1350
912c8141 1351C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1f607577 1352
954c1994 1353=cut
353c9b6f 1354
d0da05db 1355XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1356really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1357and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1358
8a58bdcf 1359Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1360patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1361
954c1994
GS
1362*/
1363
8f5283f4
KW
1364/*
1365 void below because that's the best fit, and works for Devel::PPPort
3f620621 1366=for apidoc_section $integer
545bca17 1367=for apidoc AyT||WIDEST_UTYPE
8f5283f4
KW
1368
1369Yields the widest unsigned integer type on the platform, currently either
326c768d 1370C<U32> or C<U64>. This can be used in declarations such as
8f5283f4
KW
1371
1372 WIDEST_UTYPE my_uv;
1373
1374or casts
1375
1376 my_uv = (WIDEST_UTYPE) val;
1377
1378=cut
1379
1380*/
de9e2639
KW
1381#ifdef QUADKIND
1382# define WIDEST_UTYPE U64
7c062697
KW
1383#else
1384# define WIDEST_UTYPE U32
1385#endif
1386
3912bc88
KW
1387/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1388 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1389 * bits of information are lost even on a 64-bit machine, but to get the
1390 * compiler to optimize it out if possible. This is because Configure makes
1391 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1392 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1393 *
1394 * For Coverity, be always true, because otherwise Coverity thinks
1395 * it finds several expressions that are always true, independent
1396 * of operands. Well, they are, but that is kind of the point.
220c71bf 1397 */
7e75d1a1 1398#ifndef __COVERITY__
6c5b02ac
KW
1399 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1400 * pointer) */
1401#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1402 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1403#else
1404#define FITS_IN_8_BITS(c) (1)
1405#endif
cf301eb7 1406
45f4bb73 1407/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1408 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1409 * needed. (The NV casts stop any warnings about comparison always being true
1410 * if called with an unsigned. The cast preserves the sign, which is all we
1411 * care about.) */
92a0bb24
KW
1412#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1413 __ASSERT_((NV) (n) >= 0) \
1414 withinCOUNT_KNOWN_VALID_((c), (l), (n)))
1415
1416/* For internal use only, this can be used in places where it is known that the
1417 * parameters to withinCOUNT() are valid, to avoid the asserts. For example,
1418 * inRANGE() below, calls this several times, but does all the necessary
1419 * asserts itself, once. The reason that this is necessary is that the
1420 * duplicate asserts were exceeding the internal limits of some compilers */
1421#define withinCOUNT_KNOWN_VALID_(c, l, n) \
1422 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0))))
833b0f46 1423
94250c4f
KW
1424/* Returns true if c is in the range l..u, where 'l' is non-negative
1425 * Written this way so that after optimization, only one conditional test is
4758c20d 1426 * needed. */
92a0bb24
KW
1427#define inRANGE(c, l, u) (__ASSERT_((NV) (l) >= 0) __ASSERT_((u) >= (l)) \
1428 ( (sizeof(c) == sizeof(U8)) ? inRANGE_helper_(U8, (c), (l), ((u))) \
1442da54 1429 : (sizeof(c) == sizeof(U16)) ? inRANGE_helper_(U16,(c), (l), ((u))) \
92a0bb24
KW
1430 : (sizeof(c) == sizeof(U32)) ? inRANGE_helper_(U32,(c), (l), ((u))) \
1431 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1432 inRANGE_helper_(WIDEST_UTYPE,(c), (l), ((u))))))
1433
1434/* For internal use, this is used by machine-generated code which generates
1435 * known valid calls, with a known sizeof(). This avoids the extra code and
1436 * asserts that were exceeding internal limits of some compilers. */
1437#define inRANGE_helper_(cast, c, l, u) \
1438 withinCOUNT_KNOWN_VALID_(((cast) (c)), (l), ((u) - (l)))
305fe86e 1439
41f43cc2 1440#ifdef EBCDIC
b6340bd0 1441# ifndef _ALL_SOURCE
0852beac
KW
1442 /* The native libc isascii() et.al. functions return the wrong results
1443 * on at least z/OS unless this is defined. */
b6340bd0
KW
1444# error _ALL_SOURCE should probably be defined
1445# endif
41f43cc2 1446#else
0852beac
KW
1447 /* There is a simple definition of ASCII for ASCII platforms. But the
1448 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1449 * other macros below.
3f3c579d
KW
1450 *
1451 * The cast here is used instead of '(c) >= 0', because some compilers emit
1452 * a warning that that test is always true when the parameter is an
1453 * unsigned type. khw supposes that it could be written as
1454 * && ((c) == '\0' || (c) > 0)
1455 * to avoid the message, but the cast will likely avoid extra branches even
1456 * with stupid compilers.
1457 *
1458 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1459 * a pointer) */
9c903d59 1460# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1461#endif
1462
38694112
KW
1463/* Take the eight possible bit patterns of the lower 3 bits and you get the
1464 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1465 * can be ignored. If the rest match '0', we have an octal */
1466#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1467
9fb1bf9d 1468#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1469
a500dc72
KW
1470/* Character class numbers. For internal core Perl use only. The ones less
1471 * than 32 are used in PL_charclass[] and the ones up through the one that
1472 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1473 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1474 * their actual definitions are here. If that file has a name not used here,
1475 * it won't compile.
1709d539
KW
1476 *
1477 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1478 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1479 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1480 * real Posix character class, and some (small) inefficiencies in regular
1481 * expression handling would be introduced by putting it in the middle of those
1482 * that are. Also, cntrl and ascii come after the others as it may be useful
1483 * to group these which have no members that match above Latin1, (or above
1484 * ASCII in the latter case) */
1485
1709d539
KW
1486# define _CC_WORDCHAR 0 /* \w and [:word:] */
1487# define _CC_DIGIT 1 /* \d and [:digit:] */
1488# define _CC_ALPHA 2 /* [:alpha:] */
1489# define _CC_LOWER 3 /* [:lower:] */
1490# define _CC_UPPER 4 /* [:upper:] */
1491# define _CC_PUNCT 5 /* [:punct:] */
1492# define _CC_PRINT 6 /* [:print:] */
15861f94 1493# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1494# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1495# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1496# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1497# define _CC_BLANK 11 /* [:blank:] */
1498# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1499# define _CC_CNTRL 13 /* [:cntrl:] */
1500# define _CC_ASCII 14 /* [:ascii:] */
1501# define _CC_VERTSPACE 15 /* \v */
1709d539 1502
a0947d7b
KW
1503# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1504
1709d539 1505/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1506 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1507# define _CC_IDFIRST 16
1508# define _CC_CHARNAME_CONT 17
1509# define _CC_NONLATIN1_FOLD 18
1510# define _CC_NONLATIN1_SIMPLE_FOLD 19
1511# define _CC_QUOTEMETA 20
1512# define _CC_NON_FINAL_FOLD 21
1513# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1514# define _CC_BINDIGIT 23
1515# define _CC_OCTDIGIT 24
1516# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1517
1518/* This next group is only used on EBCDIC platforms, so theoretically could be
1519 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1520# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1521/* Unused: 26-30
f4cdb42c
KW
1522 * If more bits are needed, one could add a second word for non-64bit
1523 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1524 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1525 * is used only for optimization (as of this writing), and differs in the
1526 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1527 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1528 * would just cause /i regexes which match them to run less efficiently.
1529 * Similarly the EBCDIC-only bits are used just for speed, and could be
1530 * replaced by other means */
96ac0975 1531
3a371f2f
KW
1532#if defined(PERL_CORE) || defined(PERL_EXT)
1533/* An enum version of the character class numbers, to help compilers
1534 * optimize */
1535typedef enum {
3a371f2f 1536 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1537 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1538 _CC_ENUM_ASCII = _CC_ASCII,
1539 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1540 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1541 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1542 _CC_ENUM_DIGIT = _CC_DIGIT,
1543 _CC_ENUM_GRAPH = _CC_GRAPH,
1544 _CC_ENUM_LOWER = _CC_LOWER,
1545 _CC_ENUM_PRINT = _CC_PRINT,
1546 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1547 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1548 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1549 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1550 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1551 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1552} _char_class_number;
1553#endif
1554
86f72d56 1555#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1556
6635f04f 1557START_EXTERN_C
96ac0975
NC
1558# ifdef DOINIT
1559EXTCONST U32 PL_charclass[] = {
1560# include "l1_char_class_tab.h"
1561};
1562
1563# else /* ! DOINIT */
1564EXTCONST U32 PL_charclass[];
1565# endif
6635f04f 1566END_EXTERN_C
96ac0975 1567
265c1f46 1568 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1569# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1570
1571 /* For internal core Perl use only: the base macro for defining macros like
1572 * isALPHA */
ff7ecfc3 1573# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1574 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1575
f4cdb42c
KW
1576 /* The mask for the _A versions of the macros; it just adds in the bit for
1577 * ASCII. */
1578# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1579
4650c663
KW
1580 /* For internal core Perl use only: the base macro for defining macros like
1581 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1582 * the ASCII bit are present */
b7d90381
KW
1583# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1584 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1585 == _CC_mask_A(classnum)))
f4cdb42c 1586
26c1d9d8
KW
1587/* On ASCII platforms certain classes form a single range. It's faster to
1588 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1589# ifdef EBCDIC
1590# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1591# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1592# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1593# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1594# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1595# else
26c1d9d8 1596 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1597# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1598# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1599# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1600# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1601# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1602# endif
15861f94 1603# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1604# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1605# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1606# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1607# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1608# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1609# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1610# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1611 */
d95f8b6a 1612# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1613# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1614# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1615# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1616
1617 /* continuation character for legal NAME in \N{NAME} */
1618# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1619
1620# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1621# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1622# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1623# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1624# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1625# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1626# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1627# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1628# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1629# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1630
0852beac
KW
1631# ifdef EBCDIC
1632# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1633# endif
1634
f12c0118 1635 /* Participates in a single-character fold with a character above 255 */
c62fdeb7
KW
1636# if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
1637# define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(c) \
1638 (( ! cBOOL(FITS_IN_8_BITS(c))) \
1639 || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_SIMPLE_FOLD)))
1640
1641# define IS_NON_FINAL_FOLD(c) _generic_isCC(c, _CC_NON_FINAL_FOLD)
1642# define IS_IN_SOME_FOLD_L1(c) _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1643# endif
f12c0118
KW
1644
1645 /* Like the above, but also can be part of a multi-char fold */
c62fdeb7
KW
1646# define HAS_NONLATIN1_FOLD_CLOSURE(c) \
1647 ( (! cBOOL(FITS_IN_8_BITS(c))) \
1648 || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_FOLD)))
430b7c70 1649
4eeeb416 1650# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
5e6ebb12
KW
1651
1652/* is c a control character for which we have a mnemonic? */
1653# if defined(PERL_CORE) || defined(PERL_EXT)
1654# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1655# endif
687c8d01 1656#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1657
1658 /* If we don't have perl.h, we are compiling a utility program. Below we
1659 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1660 * to it. Most are coded based on first principles. These are written to
74665a89 1661 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1662# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1663# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1664# define isSPACE_A(c) (isBLANK_A(c) \
1665 || (c) == '\n' \
1666 || (c) == '\r' \
1667 || (c) == '\v' \
0852beac 1668 || (c) == '\f')
74665a89
KW
1669 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1670 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1671 * performance (if optimization isn't on), and allow the same code to be
1672 * used for both platform types */
182c4ace
KW
1673# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1674 || inRANGE((c), 'j', 'r') \
1675 || inRANGE((c), 's', 'z')
1676# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1677 || inRANGE((c), 'J', 'R') \
1678 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1679# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1680# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1681# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1682# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1683# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1684 || inRANGE((c), 'a', 'f') \
1685 || inRANGE((c), 'A', 'F')
74665a89
KW
1686# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1687 || (c) == '#' || (c) == '$' || (c) == '%' \
1688 || (c) == '&' || (c) == '\'' || (c) == '(' \
1689 || (c) == ')' || (c) == '*' || (c) == '+' \
1690 || (c) == ',' || (c) == '.' || (c) == '/' \
1691 || (c) == ':' || (c) == ';' || (c) == '<' \
1692 || (c) == '=' || (c) == '>' || (c) == '?' \
1693 || (c) == '@' || (c) == '[' || (c) == '\\' \
1694 || (c) == ']' || (c) == '^' || (c) == '_' \
1695 || (c) == '`' || (c) == '{' || (c) == '|' \
1696 || (c) == '}' || (c) == '~')
1697# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1698# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1699
0852beac 1700# ifdef EBCDIC
74665a89
KW
1701 /* The below is accurate for the 3 EBCDIC code pages traditionally
1702 * supported by perl. The only difference between them in the controls
1703 * is the position of \n, and that is represented symbolically below */
1704# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1705 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1706 || (c) == '\t' || (c) == '\v' \
182c4ace 1707 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1708 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1709 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1710 DC[1-3] */ \
74665a89
KW
1711 || (c) == 0x18 /* U+18 CAN */ \
1712 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1713 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1714 || (c) == 0x26 /* U+17 ETB */ \
1715 || (c) == 0x27 /* U+1B ESC */ \
1716 || (c) == 0x2D /* U+05 ENQ */ \
1717 || (c) == 0x2E /* U+06 ACK */ \
1718 || (c) == 0x32 /* U+16 SYN */ \
1719 || (c) == 0x37 /* U+04 EOT */ \
1720 || (c) == 0x3C /* U+14 DC4 */ \
1721 || (c) == 0x3D /* U+15 NAK */ \
1722 || (c) == 0x3F)/* U+1A SUB */
0852beac 1723# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1724# else /* isASCII is already defined for ASCII platforms, so can use that to
1725 define isCNTRL */
1726# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1727# endif
1728
3ffc8c70 1729 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1730 * during debugging, and it seems best to keep them. We may be called
1731 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1732 * do anything anyway, so make it not a problem */
1733# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1734# define NATIVE_TO_LATIN1(ch) (ch)
1735# endif
3ded5eb0
KW
1736# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1737# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1738# define isBLANK_L1(c) (isBLANK_A(c) \
1739 || (FITS_IN_8_BITS(c) \
1740 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1741# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1742# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1743# define isLOWER_L1(c) (isLOWER_A(c) \
1744 || (FITS_IN_8_BITS(c) \
ae683a5f 1745 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1746 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1747 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1748 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1749 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1750# define isPRINT_L1(c) (isPRINT_A(c) \
1751 || (FITS_IN_8_BITS(c) \
1752 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1753# define isPUNCT_L1(c) (isPUNCT_A(c) \
1754 || (FITS_IN_8_BITS(c) \
ae683a5f 1755 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1756 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1757 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1758 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1759 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1760 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1761 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1762# define isSPACE_L1(c) (isSPACE_A(c) \
1763 || (FITS_IN_8_BITS(c) \
ae683a5f 1764 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1765 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1766# define isUPPER_L1(c) (isUPPER_A(c) \
1767 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1768 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1769 0xC0, 0xDE) \
3ded5eb0
KW
1770 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1771# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1772# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1773# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1774 || isBLANK_L1(c) \
1775 || (c) == '-' \
1776 || (c) == '(' \
1777 || (c) == ')')
1778 /* The following are not fully accurate in the above-ASCII range. I (khw)
1779 * don't think it's necessary to be so for the purposes where this gets
1780 * compiled */
1781# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1782# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1783
1784 /* And these aren't accurate at all. They are useful only for above
1785 * Latin1, which utilities and bootstrapping don't deal with */
1786# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1787# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1788# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1789
1790 /* Many of the macros later in this file are defined in terms of these. By
1791 * implementing them with a function, which converts the class number into
1792 * a call to the desired macro, all of the later ones work. However, that
1793 * function won't be actually defined when building a utility program (no
1794 * perl.h), and so a compiler error will be generated if one is attempted
1795 * to be used. And the above-Latin1 code points require Unicode tables to
1796 * be present, something unlikely to be the case when bootstrapping */
1797# define _generic_isCC(c, classnum) \
1798 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1799# define _generic_isCC_A(c, classnum) \
1800 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1801#endif /* End of no perl.h H_PERL */
8a58bdcf 1802
e66b99e9
KW
1803#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1804#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1805#define isASCII_A(c) isASCII(c)
1806#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1807#define isBLANK(c) isBLANK_A(c)
1808#define isCNTRL(c) isCNTRL_A(c)
1809#define isDIGIT(c) isDIGIT_A(c)
1810#define isGRAPH(c) isGRAPH_A(c)
1811#define isIDFIRST(c) isIDFIRST_A(c)
1812#define isLOWER(c) isLOWER_A(c)
1813#define isPRINT(c) isPRINT_A(c)
779cf272 1814#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1815#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1816#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1817#define isPUNCT(c) isPUNCT_A(c)
1818#define isSPACE(c) isSPACE_A(c)
1819#define isUPPER(c) isUPPER_A(c)
1820#define isWORDCHAR(c) isWORDCHAR_A(c)
1821#define isXDIGIT(c) isXDIGIT_A(c)
1822
1823/* ASCII casing. These could also be written as
1824 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1825 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1826 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1827 work because the _MOD does not apply in the ASCII range).
1828
1829 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1830 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1831 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1832#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1833#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1834
25200305
KW
1835/* In the ASCII range, these are equivalent to what they're here defined to be.
1836 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1837 * this detail. Actually this works for all UTF-8 invariants, not just the
1838 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1839#define toFOLD(c) toLOWER(c)
25200305
KW
1840#define toTITLE(c) toUPPER(c)
1841
c753c8d3
KW
1842#define toLOWER_A(c) toLOWER(c)
1843#define toUPPER_A(c) toUPPER(c)
25200305
KW
1844#define toFOLD_A(c) toFOLD(c)
1845#define toTITLE_A(c) toTITLE(c)
1a0901db 1846
4650c663 1847/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1848#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1849 ? (c) \
f4cd282c 1850 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1851#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1852
1a0901db 1853/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1854 * all mapped to one of them, and these need special handling; returns the
1855 * input itself if is out-of-range */
b2bf251f 1856#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1857 ? (c) \
f4cd282c 1858 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1859#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1860
beab9ebe
KW
1861/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1862
1863/* For internal core Perl use only: the base macro for defining macros like
1864 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1865 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1866 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1867 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1868 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1869 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1870 * code points under locale rules, use the macros based on _generic_LC_uvchr
1871 * instead of this. */
beab9ebe
KW
1872#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1873 (! FITS_IN_8_BITS(c) \
1874 ? 0 \
31f05a37
KW
1875 : IN_UTF8_CTYPE_LOCALE \
1876 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1877 : cBOOL(non_utf8))
1878
1879/* For internal core Perl use only: a helper macro for defining macros like
1880 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1881 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1882 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1883 * _generic_LC_base for more info */
1884#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1885 _generic_LC_base(c,utf8_locale_classnum, \
1886 non_utf8_func( (_LC_CAST) (c)))
1887
1888/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1889 * 'c' is the platform's native underscore character */
1890#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1891 _generic_LC_base(c, utf8_locale_classnum, \
1892 (non_utf8_func( (_LC_CAST) (c)) \
1893 || (char)(c) == '_'))
1894
1895/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1896 * helper macros. The reason for using the PL_latin arrays is in case the
1897 * system function is defective; it ensures uniform results that conform to the
b257a28c 1898 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1899#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1900 ? (c) \
31f05a37
KW
1901 : (IN_UTF8_CTYPE_LOCALE) \
1902 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1903 : (cast)function((cast)(c)))
beab9ebe 1904
31f05a37
KW
1905/* Note that the result can be larger than a byte in a UTF-8 locale. It
1906 * returns a single value, so can't adequately return the upper case of LATIN
1907 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1908 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1909 * otherwise returns its input. It does not handle the anomalies in UTF-8
1910 * Turkic locales. */
beab9ebe
KW
1911#define _generic_toUPPER_LC(c, function, cast) \
1912 (! FITS_IN_8_BITS(c) \
1913 ? (c) \
31f05a37 1914 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1915 ? (cast)function((cast)(c)) \
31f05a37
KW
1916 : ((((U8)(c)) == MICRO_SIGN) \
1917 ? GREEK_CAPITAL_LETTER_MU \
1918 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1919 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1920 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1921 ? (__ASSERT_(0) (c)) \
1922 : PL_mod_latin1_uc[ (U8) (c) ])))))
1923
1924/* Note that the result can be larger than a byte in a UTF-8 locale. It
1925 * returns a single value, so can't adequately return the fold case of LATIN
1926 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1927 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1928 * otherwise returns its input. It does not handle the anomalies in UTF-8
1929 * Turkic locales */
beab9ebe 1930#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1931 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1932 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1933 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1934 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1935 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1936
84061b6a 1937/* Use the libc versions for these if available. */
f05550c0 1938#if defined(HAS_ISASCII)
84061b6a
KW
1939# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1940#else
1941# define isASCII_LC(c) isASCII(c)
1942#endif
1943
f05550c0 1944#if defined(HAS_ISBLANK)
84061b6a 1945# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1946#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1947# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1948#endif
1949
f05550c0 1950#define _LC_CAST U8
bbce6d69 1951
f05550c0 1952#ifdef WIN32
375f5f06
KW
1953 /* The Windows functions don't bother to follow the POSIX standard, which
1954 * for example says that something can't both be a printable and a control.
1955 * But Windows treats the \t control as a printable, and does such things
1956 * as making superscripts into both digits and punctuation. This tames
1957 * these flaws by assuming that the definitions of both controls and space
1958 * are correct, and then making sure that other definitions don't have
1959 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1960 * Not all possible weirdnesses are checked for, just the ones that were
1961 * detected on actual Microsoft code pages */
1962
b7d90381
KW
1963# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1964# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1965
1966# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1967 && isALPHANUMERIC_LC(c))
1968# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1969 ! isPUNCT_LC(c))
1970# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1971 isALPHANUMERIC_LC(c))
1972# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1973# define isIDFIRST_LC(c) (((c) == '_') \
1974 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1975# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1976# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1977# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1978# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1979# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1980# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1981 && isALPHANUMERIC_LC(c))
f05550c0
BF
1982
1983# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1984# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1985# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1986
1987#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1988 /* For most other platforms */
beab9ebe 1989
f05550c0
BF
1990# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1991# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1992# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1993# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1994# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1995# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1996# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1997# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1998# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1999# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
2000# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
2001# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
2002# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
2003
2004
2005# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
2006# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
2007# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
2008
2009#else /* The final fallback position */
2010
b7d90381
KW
2011# define isALPHA_LC(c) (isascii(c) && isalpha(c))
2012# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
2013# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
2014# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
2015# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 2016# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
2017# define isLOWER_LC(c) (isascii(c) && islower(c))
2018# define isPRINT_LC(c) (isascii(c) && isprint(c))
2019# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
2020# define isSPACE_LC(c) (isascii(c) && isspace(c))
2021# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 2022# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 2023# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
2024
2025# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
2026# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
2027# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 2028
f05550c0 2029#endif
55204971 2030
eba68aa0
KW
2031#define isIDCONT(c) isWORDCHAR(c)
2032#define isIDCONT_A(c) isWORDCHAR_A(c)
2033#define isIDCONT_L1(c) isWORDCHAR_L1(c)
2034#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 2035#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 2036
4650c663 2037/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
2038 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
2039 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
2040 * classes where there is a macro or function 'above_latin1' that takes the
2041 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
2042 * classes which have simple definitions, avoiding the overhead of an inversion
2043 * list binary search. _generic_invlist_uvchr() can be used
4650c663 2044 * for classes where that overhead is faster than a direct lookup.
d0da05db 2045 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
2046 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2047 * have duplicate checks here, so could create versions of the macros that
2048 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
2049
2050/* Note that all ignore 'use bytes' */
1e222e4f
KW
2051#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2052 ? _generic_isCC(c, classnum) \
cd500f2f 2053 : above_latin1(c))
2366ba44 2054#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 2055 ? _generic_isCC(c, classnum) \
922e8cb4 2056 : _is_uni_FOO(classnum, c))
2366ba44
KW
2057#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2058#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
2059#define isASCII_uvchr(c) isASCII(c)
2060#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2061#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
2062#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2063#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
2064#define isIDCONT_uvchr(c) \
2065 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2066#define isIDFIRST_uvchr(c) \
2067 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
2068#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2069#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 2070
2366ba44 2071#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
2072#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2073#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2074
2366ba44 2075#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 2076#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 2077#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
2078#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2079
2080#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2081#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2082#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2083#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2084
2085/* For backwards compatibility, even though '_uni' should mean official Unicode
2086 * code points, in Perl it means native for those below 256 */
2087#define isALPHA_uni(c) isALPHA_uvchr(c)
2088#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2089#define isASCII_uni(c) isASCII_uvchr(c)
2090#define isBLANK_uni(c) isBLANK_uvchr(c)
2091#define isCNTRL_uni(c) isCNTRL_uvchr(c)
2092#define isDIGIT_uni(c) isDIGIT_uvchr(c)
2093#define isGRAPH_uni(c) isGRAPH_uvchr(c)
2094#define isIDCONT_uni(c) isIDCONT_uvchr(c)
2095#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2096#define isLOWER_uni(c) isLOWER_uvchr(c)
2097#define isPRINT_uni(c) isPRINT_uvchr(c)
2098#define isPUNCT_uni(c) isPUNCT_uvchr(c)
2099#define isSPACE_uni(c) isSPACE_uvchr(c)
2100#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2101#define isUPPER_uni(c) isUPPER_uvchr(c)
2102#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2103#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2104#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2105#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2106#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2107#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2108#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2109
4650c663
KW
2110/* For internal core Perl use only: the base macros for defining macros like
2111 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2112 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2113 * simple class definitions; the other for more complex. These are like
d0da05db 2114 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2115#define _generic_LC_uvchr(latin1, above_latin1, c) \
2116 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2117#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2118 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2119
2366ba44
KW
2120#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2121#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2122 _CC_ALPHANUMERIC, c)
b7d90381
KW
2123#define isASCII_LC_uvchr(c) isASCII_LC(c)
2124#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2125 is_HORIZWS_cp_high, c)
feeab5a9 2126#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2127#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2128#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2129#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2130 _is_uni_perl_idcont, c)
b7d90381 2131#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2132 _is_uni_perl_idstart, c)
2366ba44
KW
2133#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2134#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2135#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2136#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2137#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2138 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2139#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2140#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2141 _CC_WORDCHAR, c)
b7d90381
KW
2142#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2143 is_XDIGIT_cp_high, c)
e712593e 2144
b7d90381 2145#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2146
da8c1a98
KW
2147/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2148 * they don't otherwise go out of their way to look for malformed UTF-8. If
2149 * they can return accurate results without knowing if the input is otherwise
2150 * malformed, they do so. For example isASCII is accurate in spite of any
2151 * non-length malformations because it looks only at a single byte. Likewise
2152 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2153 * variant ones return FALSE. But, if the input has to be well-formed in order
2154 * for the results to be accurate, the macros will test and if malformed will
2155 * call a routine to die
2156 *
2157 * Except for toke.c, the macros do assume that e > p, asserting that on
2158 * DEBUGGING builds. Much code that calls these depends on this being true,
2159 * for other reasons. toke.c is treated specially as using the regular
2160 * assertion breaks it in many ways. All strings that these operate on there
2161 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2162 * bunch of code in toke.c assumes that this is true, so the assertion allows
2163 * for that */
2164#ifdef PERL_IN_TOKE_C
2165# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2166#else
2167# define _utf8_safe_assert(p,e) ((e) > (p))
2168#endif
2169
2170#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2171 ((! _utf8_safe_assert(p, e)) \
2172 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2173 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2174 ? _generic_isCC(*(p), classnum) \
2175 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2176 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2177 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2178 classnum) \
2179 : (_force_out_malformed_utf8_message( \
2180 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2181 : above_latin1))
b7d90381
KW
2182/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2183 * 'above_latin1' can be a macro */
da8c1a98
KW
2184#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2185 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2186#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2187 _generic_utf8_safe(classnum, p, e, \
2188 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2189 ? (_force_out_malformed_utf8_message( \
2190 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2191 : above_latin1(p)))
2366ba44
KW
2192/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2193 * 'above_latin1' parameter */
2194#define _generic_invlist_utf8_safe(classnum, p, e) \
2195 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2196
cc8ab7c0 2197/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2198 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2199 * class is TRUE for. Hence it can skip the tests for this range.
2200 * 'above_latin1' should include its arguments */
da8c1a98
KW
2201#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2202 (__ASSERT_(_utf8_safe_assert(p, e)) \
2203 (UTF8_IS_INVARIANT(*(p))) \
2204 ? _generic_isCC(*(p), classnum) \
2205 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2206 ? 0 /* Note that doesn't check validity for latin1 */ \
2207 : above_latin1)
2208
84238efa 2209
059703b0
KW
2210#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2211#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2212#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2213#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2214#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2215#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2216#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2217#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2218#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2219#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2220#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2221#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2222#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2223#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2224#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2225#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2226#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2227#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2228
2366ba44 2229#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2230#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2231 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2232#define isASCII_utf8_safe(p, e) \
2233 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2234 * works */ \
2235 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2236#define isBLANK_utf8_safe(p, e) \
2366ba44 2237 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2238
e8fa43e2
KW
2239#ifdef EBCDIC
2240 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2241 * more efficient macro instead of the more general one */
da8c1a98 2242# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2243 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2244#else
da8c1a98 2245# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2246#endif
2247
da8c1a98
KW
2248#define isDIGIT_utf8_safe(p, e) \
2249 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2250 _is_utf8_FOO(_CC_DIGIT, p, e))
2251#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2252#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2253 _is_utf8_perl_idcont, p, e)
e5dcd934 2254
c11ff943
KW
2255/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2256 * IDFIRST is an alnum. See
8034715d 2257 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2258 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2259 * This used to be not the XID version, but we decided to go with the more
2260 * modern Unicode definition */
da8c1a98
KW
2261#define isIDFIRST_utf8_safe(p, e) \
2262 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2263 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2264
2366ba44
KW
2265#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2266#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2267#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2268#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2269#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2270 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2271#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2272#define isVERTWS_utf8_safe(p, e) \
2366ba44 2273 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2274#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2275 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2276#define isXDIGIT_utf8_safe(p, e) \
2277 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2278 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2279 ? (_force_out_malformed_utf8_message( \
2280 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2281 : is_XDIGIT_high(p)))
a0ed51b3 2282
059703b0
KW
2283#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2284#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2285#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2286#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2287
567b353c 2288/* For internal core use only, subject to change */
059703b0
KW
2289#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2290#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2291#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2292#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2293
2294#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2295#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2296#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2297#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2298
059703b0
KW
2299#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2300#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2301#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2302#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2303#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2304#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2305#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2306#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2307#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2308#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2309#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2310#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2311#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2312#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2313#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2314#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2315#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2316
da8c1a98
KW
2317/* For internal core Perl use only: the base macros for defining macros like
2318 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2319 * point in 'p' is within the 0-255 range, it uses locale rules from the
2320 * passed-in 'macro' parameter */
2321#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2322 (__ASSERT_(_utf8_safe_assert(p, e)) \
2323 (UTF8_IS_INVARIANT(*(p))) \
2324 ? macro(*(p)) \
2325 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2326 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2327 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2328 : (_force_out_malformed_utf8_message( \
2329 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2330 : above_latin1))
2331
2366ba44
KW
2332#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2333 _generic_LC_utf8_safe(macro, p, e, \
2334 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2335
2336#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2337 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2338
2366ba44 2339#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2340 _generic_LC_utf8_safe(classnum, p, e, \
2341 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2342 ? (_force_out_malformed_utf8_message( \
2343 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2344 : above_latin1(p)))
2345
2346#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2347 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2348 _CC_ALPHANUMERIC, p, e)
2349#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2350 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2351#define isASCII_LC_utf8_safe(p, e) \
2352 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2353#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2354 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2355#define isCNTRL_LC_utf8_safe(p, e) \
2356 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2357#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2358 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2359#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2360 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2361#define isIDCONT_LC_utf8_safe(p, e) \
2362 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2363 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2364#define isIDFIRST_LC_utf8_safe(p, e) \
2365 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2366 _is_utf8_perl_idstart, p, e)
da8c1a98 2367#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2368 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2369#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2370 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2371#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2372#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2373 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2374#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2375 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2376#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2377 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2378#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2379 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2380#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2381 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2382
fbc19f27
KW
2383/* Macros for backwards compatibility and for completeness when the ASCII and
2384 * Latin1 values are identical */
b7d90381
KW
2385#define isALPHAU(c) isALPHA_L1(c)
2386#define isDIGIT_L1(c) isDIGIT_A(c)
2387#define isOCTAL(c) isOCTAL_A(c)
2388#define isOCTAL_L1(c) isOCTAL_A(c)
2389#define isXDIGIT_L1(c) isXDIGIT_A(c)
2390#define isALNUM(c) isWORDCHAR(c)
a377c856 2391#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2392#define isALNUMU(c) isWORDCHAR_L1(c)
2393#define isALNUM_LC(c) isWORDCHAR_LC(c)
2394#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2395#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2396#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2397#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2398#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2399#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2400#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2401#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2402#define isALNUMC(c) isALPHANUMERIC(c)
2403#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2404#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2405#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2406#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2407#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2408#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2409
2bd1cbf6
KW
2410/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2411 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2412 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2413 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2414 * the outlier from the block that contains the other controls, just like
2415 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2416 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2417 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2418 * etc. */
2bd1cbf6 2419#ifndef EBCDIC
75763b3a 2420# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2421#else
75763b3a
KW
2422# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2423 ((isPRINT_A(c)) \
2424 ? (UNLIKELY((c) == '?') \
2425 ? QUESTION_MARK_CTRL \
2426 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2427 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2428 ? '?' \
2429 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2430#endif
bbce6d69 2431
837781cc
KW
2432/*
2433=for apidoc Ay||line_t
2434The typedef to use to declare variables that are to hold line numbers.
2435
2436=cut
2437
2438 Line numbers are unsigned, 32 bits.
2439*/
dea28490 2440typedef U32 line_t;
e5dcd934 2441#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2442
91152fc1
DG
2443/* Helpful alias for version prescan */
2444#define is_LAX_VERSION(a,b) \
1604cfb0 2445 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
91152fc1
DG
2446
2447#define is_STRICT_VERSION(a,b) \
1604cfb0 2448 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
91152fc1
DG
2449
2450#define BADVERSION(a,b,c) \
1604cfb0
MS
2451 if (b) { \
2452 *b = c; \
2453 } \
2454 return a;
8c52afec 2455
1ce77b7d
KW
2456/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2457 * a-f) to its numeric value without using any branches. The input is
2458 * validated only by an assert() in DEBUGGING builds.
2459 *
2460 * It works by right shifting and isolating the bit that is 0 for the digits,
2461 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2462 * position, and then to the eights position. Both are added together to form
2463 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2464 * final four bits of the input to form the correct value. */
2465#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2466 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2467 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2468 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2469
2470/* The argument is a string pointer, which is advanced. */
2471#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2472
cb27eebd
KW
2473/* Converts a character known to represent an octal digit (0-7) to its numeric
2474 * value. The input is validated only by an assert() in DEBUGGING builds. In
2475 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2476#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2477
305b8651 2478/* Efficiently returns a boolean as to if two native characters are equivalent
f1460a66 2479 * case-insensitively. At least one of the characters must be one of [A-Za-z];
305b8651
KW
2480 * the ALPHA in the name is to remind you of that. This is asserted() in
2481 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2482 * works (on valid input) for both non- and UTF-8-encoded bytes.
2483 *
2484 * When one of the inputs is a compile-time constant and gets folded by the
2485 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2486 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2487 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2488 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2489 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2490 * just a single 0, in the bit position where the upper- and lowercase differ.
2491 * */
305b8651
KW
2492#define isALPHA_FOLD_EQ(c1, c2) \
2493 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2494 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2495#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2496
8e84507e 2497/*
3f620621 2498=for apidoc_section $memory
ccfc67b7 2499
a02a5408 2500=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2501The XSUB-writer's interface to the C C<malloc> function.
2502
596f7718 2503Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2504
c5008215
JC
2505In 5.9.3, Newx() and friends replace the older New() API, and drops
2506the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2507themselves. This aid has been superseded by a new build option,
d10b4965 2508PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2509there for use in XS modules supporting older perls.
2510
a02a5408 2511=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2512The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2513cast. See also C<L</Newx>>.
954c1994 2514
596f7718 2515Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2516
a02a5408 2517=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2518The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2519memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2520
596f7718 2521Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2522
954c1994
GS
2523=for apidoc Am|void|Renew|void* ptr|int nitems|type
2524The XSUB-writer's interface to the C C<realloc> function.
2525
596f7718 2526Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2527
954c1994
GS
2528=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2529The XSUB-writer's interface to the C C<realloc> function, with
2530cast.
2531
596f7718 2532Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2533
49b8b560 2534=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2535The XSUB-writer's interface to the C C<free> function.
2536
596f7718 2537This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2538
3f620621 2539=for apidoc_section $string
954c1994
GS
2540=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2541The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2542source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2543C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2544
e90e2364 2545=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2546Like C<Move> but returns C<dest>. Useful
72d33970 2547for encouraging compilers to tail-call
e90e2364
NC
2548optimise.
2549
954c1994
GS
2550=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2551The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2552source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2553C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2554
e90e2364
NC
2555=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2556
796b6530 2557Like C<Copy> but returns C<dest>. Useful
72d33970 2558for encouraging compilers to tail-call
e90e2364
NC
2559optimise.
2560
954c1994
GS
2561=for apidoc Am|void|Zero|void* dest|int nitems|type
2562
2563The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2564destination, C<nitems> is the number of items, and C<type> is the type.
2565
e90e2364
NC
2566=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2567
72d33970
FC
2568Like C<Zero> but returns dest. Useful
2569for encouraging compilers to tail-call
e90e2364
NC
2570optimise.
2571
3f620621 2572=for apidoc_section $utility
e538104b 2573=for apidoc Amu|void|StructCopy|type *src|type *dest|type
4375e838 2574This is an architecture-independent macro to copy one structure to another.
954c1994 2575
7e337ee0
JH
2576=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2577
2578Fill up memory with a byte pattern (a byte repeated over and over
2579again) that hopefully catches attempts to access uninitialized memory.
2580
2581=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2582
2583PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2584
1c12ffb4 2585=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2586
2587PoisonWith(0xEF) for catching access to freed memory.
2588
9965345d
JH
2589=for apidoc Am|void|Poison|void* dest|int nitems|type
2590
7e337ee0 2591PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2592
2593=cut */
954c1994 2594
561b68a9
SH
2595/* Maintained for backwards-compatibility only. Use newSV() instead. */
2596#ifndef PERL_CORE
ff06c60c 2597#define NEWSV(x,len) newSV(len)
561b68a9 2598#endif
ff06c60c 2599
b7112dce 2600#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2601
a500027b 2602#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2603
27d5b266 2604#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2605
2606/* This expression will be constant-folded at compile time. It checks
2607 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2608 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2609 * As well as avoiding the need for a run-time check in some cases, it's
2610 * designed to avoid compiler warnings like:
2611 * comparison is always false due to limited range of data type
73e8ff00
DM
2612 * It's mathematically equivalent to
2613 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2614 */
2615
2616# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2617 ( sizeof(MEM_SIZE) < sizeof(n) \
2618 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2619
88f9f128 2620/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2621 * compiler warnings. We *want* to write the expression as
2622 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2623 * (for some compile-time constant C), but even when the LHS
2624 * constant-folds to false at compile-time, g++ insists on emitting
2625 * warnings about the RHS (e.g. "comparison is always false"), so instead
2626 * we write it as
e6bdf523 2627 *
d98e5cde 2628 * (cond ? n : X) > C
88f9f128 2629 *
d98e5cde
DM
2630 * where X is a constant with X > C always false. Choosing a value for X
2631 * is tricky. If 0, some compilers will complain about 0 > C always being
2632 * false; if 1, Coverity complains when n happens to be the constant value
2633 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2634 * for X and hope that nothing else whines.
e6bdf523
DM
2635 */
2636
2637# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2638 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2639 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2640
2641# define MEM_WRAP_CHECK(n,t) \
1604cfb0 2642 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
e6bdf523
DM
2643 && (croak_memory_wrap(),0))
2644
2645# define MEM_WRAP_CHECK_1(n,t,a) \
1604cfb0
MS
2646 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2647 && (Perl_croak_nocontext("%s",(a)),0))
e6bdf523 2648
814eedc8
DD
2649/* "a" arg must be a string literal */
2650# define MEM_WRAP_CHECK_s(n,t,a) \
1604cfb0
MS
2651 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2652 && (Perl_croak_nocontext("" a ""),0))
814eedc8 2653
8b44ba4c 2654#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2655
a500027b 2656#define PERL_STRLEN_ROUNDUP(n) ((void)(((n) > MEM_SIZE_MAX - 2 * PERL_STRLEN_ROUNDUP_QUANTUM) ? (croak_memory_wrap(),0) : 0), _PERL_STRLEN_ROUNDUP_UNCHECKED(n))
27d5b266
JH
2657#else
2658
410319be
NC
2659#define MEM_WRAP_CHECK(n,t)
2660#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2661#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2662#define MEM_WRAP_CHECK_(n,t)
2663
a500027b 2664#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2665
1936d2a7 2666#endif
8b44ba4c 2667
fe4f188c 2668#ifdef PERL_MEM_LOG
46c6c7e2 2669/*
9f653bb5 2670 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2671 * go through functions, which are handy for debugging breakpoints, but
2672 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2673 * line number, and C function name if available) passed in. This info can
2674 * then be used for logging the calls, for which one gets a sample
73d1d973 2675 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2676 *
46c6c7e2 2677 * Known problems:
94e892a6 2678 * - not all memory allocs get logged, only those
46c6c7e2 2679 * that go through Newx() and derivatives (while all
94e892a6 2680 * Safefrees do get logged)
46c6c7e2
JH
2681 * - __FILE__ and __LINE__ do not work everywhere
2682 * - __func__ or __FUNCTION__ even less so
2683 * - I think more goes on after the perlio frees but
2684 * the thing is that STDERR gets closed (as do all
2685 * the file descriptors)
2686 * - no deeper calling stack than the caller of the Newx()
2687 * or the kind, but do I look like a C reflection/introspection
2688 * utility to you?
2689 * - the function prototypes for the logging functions
2690 * probably should maybe be somewhere else than handy.h
2691 * - one could consider inlining (macrofying) the logging
2692 * for speed, but I am too lazy
2693 * - one could imagine recording the allocations in a hash,
2694 * (keyed by the allocation address?), and maintain that
2695 * through reallocs and frees, but how to do that without
2696 * any News() happening...?
73d1d973 2697 * - lots of -Ddefines to get useful/controllable output
b953482e 2698 * - lots of ENV reads
46c6c7e2
JH
2699 */
2700
0b0ab801 2701# ifdef PERL_CORE
73d1d973 2702# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2703enum mem_log_type {
2704 MLT_ALLOC,
2705 MLT_REALLOC,
d7a2c63c
MHM
2706 MLT_FREE,
2707 MLT_NEW_SV,
2708 MLT_DEL_SV
0b0ab801
MHM
2709};
2710# endif
12754f92 2711# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2712void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2713void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2714# endif
0b0ab801
MHM
2715# endif
2716
fe4f188c
JH
2717#endif
2718
2719#ifdef PERL_MEM_LOG
d1401ee9
MHM
2720#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2721#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2722#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2723#endif
2724
2725#ifndef MEM_LOG_ALLOC
2726#define MEM_LOG_ALLOC(n,t,a) (a)
2727#endif
2728#ifndef MEM_LOG_REALLOC
2729#define MEM_LOG_REALLOC(n,t,v,a) (a)
2730#endif
2731#ifndef MEM_LOG_FREE
2732#define MEM_LOG_FREE(a) (a)
2733#endif
2734
d1401ee9
MHM
2735#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2736#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2737#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2738
2739#ifndef PERL_CORE
a02a5408
JC
2740/* pre 5.9.x compatibility */
2741#define New(x,v,n,t) Newx(v,n,t)
2742#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2743#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2744#endif
a02a5408 2745
ff68c719 2746#define Renew(v,n,t) \
1604cfb0 2747 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2748#define Renewc(v,n,t,c) \
1604cfb0 2749 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2750
2751#ifdef PERL_POISON
2752#define Safefree(d) \
06c0cc96 2753 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2754#else
fe4f188c 2755#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2756#endif
55497cff 2757
dbb57106
YO
2758/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2759 * as it handles cases like constant string arguments without throwing warnings *
2760 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2761#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2762
45908e4d
YO
2763
2764#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2765#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2766#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2767
bdd1531d 2768/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2769#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2770#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
45908e4d 2771#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2772
7e337ee0
JH
2773#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2774#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2775#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2776#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2777
caa674f3
DD
2778#ifdef PERL_POISON
2779# define PERL_POISON_EXPR(x) x
2780#else
2781# define PERL_POISON_EXPR(x)
2782#endif
2783
be319be3 2784/* Shallow copy */
ff68c719 2785#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2786
1b7e2294 2787/*
3f620621 2788=for apidoc_section $utility
1b7e2294
KW
2789
2790=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2791
2792Returns the number of elements in the input C array (so you want your
2793zero-based indices to be less than but not equal to).
2794
2795=for apidoc Am|void *|C_ARRAY_END|void *a
2796
2797Returns a pointer to one element past the final element of the input C array.
2798
2799=cut
2800
2801C_ARRAY_END is one past the last: half-open/half-closed range, not
2802last-inclusive range.
2803*/
622913ab 2804#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2805#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2806
2cc61e15
DD
2807#ifdef NEED_VA_COPY
2808# ifdef va_copy
2809# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2810# elif defined(__va_copy)
2811# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2812# else
07798b17 2813# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2814# endif
2815#endif
2816
472d47bc
SB
2817/* convenience debug macros */
2818#ifdef USE_ITHREADS
2819#define pTHX_FORMAT "Perl interpreter: 0x%p"
2820#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2821#define pTHX_VALUE_ (void *)my_perl,
2822#define pTHX_VALUE (void *)my_perl
2823#define pTHX__VALUE_ ,(void *)my_perl,
2824#define pTHX__VALUE ,(void *)my_perl
472d47bc 2825#else
3609ea0d 2826#define pTHX_FORMAT
472d47bc 2827#define pTHX__FORMAT
3609ea0d 2828#define pTHX_VALUE_
472d47bc 2829#define pTHX_VALUE
3609ea0d 2830#define pTHX__VALUE_
472d47bc
SB
2831#define pTHX__VALUE
2832#endif /* USE_ITHREADS */
3609ea0d 2833
2acdbac1
NC
2834/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2835 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2836 CPAN::Unpack show any users outside the core. */
2837#ifdef PERL_CORE
dc6e8de0
A
2838# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2839 "Use of " s " is deprecated")
c9680906
A
2840# define deprecate_disappears_in(when,message) \
2841 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2842 message ", and will disappear in Perl " when)
ac641426
A
2843# define deprecate_fatal_in(when,message) \
2844 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2845 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2846#endif
2847
dfff4baf
BF
2848/* Internal macros to deal with gids and uids */
2849#ifdef PERL_CORE
2850
2851# if Uid_t_size > IVSIZE
2852# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2853# define SvUID(sv) SvNV(sv)
07798b17
AC
2854# elif Uid_t_sign <= 0
2855# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2856# define SvUID(sv) SvIV(sv)
dfff4baf 2857# else
07798b17
AC
2858# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2859# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2860# endif /* Uid_t_size */
2861
2862# if Gid_t_size > IVSIZE
2863# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2864# define SvGID(sv) SvNV(sv)
07798b17
AC
2865# elif Gid_t_sign <= 0
2866# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2867# define SvGID(sv) SvIV(sv)
dfff4baf 2868# else
07798b17
AC
2869# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2870# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2871# endif /* Gid_t_size */
2872
2873#endif
2874
6a5bc5ac 2875#endif /* PERL_HANDY_H_ */
9d745869 2876
e9a8c099 2877/*
14d04a33 2878 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2879 */